What is generally considered good/bad conduct regarding sharing reports?
Journals generally dictate their own policies for sharing referee reports. In many cases, sharing the report with anyone other than the authors can be considered a breach of trust. Refer to the confidentiality clause located at the bottom of the email you received containing the referee comments. If one does exist, sharing the report publicly will put you in the wrong and the journal may decline to review future articles authored by your person.
Many journals may not have a confidentiality clause, but still it is your responsibility to check the For Authors section in the journal website before you share the reports publicly.
Therefore, always make sure you are in the clear before sharing reports publicly.
Having said this, a statement must be made regarding the the types of review that are there. There are three basic types of reviews:
- Open peer review: You know who is reviewing and they know you too. (Elife)
- Blind peer review: You don't know who is reviewing, but they know you. (Nature Comm.)
- Double blind peer review: You don't know the reviewers and vice versa (I cannot at this moment think of a journal)
I do not know of any Hitchhiker's guide to sharing peer review reports. But, if you are sharing the reports publicly, then in order of decreasing importance:
- If this information exists, remove reviewer names, affiliations and any information that would make the reviewer identifiable.
- Share the full report and quote it verbatim, this will put you in the clear if some misunderstandings arise later on.
- Include a confidentiality clause when sharing the report.
Our world is very small, it is very likely that someone within your network is your reviewer. So, when sharing the report, do so from the standpoint of an individual who is receptive to the comments.
Be receptive to comments that are within reason, not the ones which say the author should travel to the other side of the universe, gather some pixie dust, de-constitute and reconstitute it and finally compare it to their pixie dust in three technical and 100 biological replicates.
Are there any clearer guidelines or reasons why we should keep the reports private?
Look above regarding confidentiality clauses and journal statements for authors.
Is it considered a breach of confidence of the anonymous referee?
If sharing the report is not a breach of trust from the journal's standing and information in the report does not clearly identify the reviewer, then no I do not think that is a breach of confidence. But, if I were the reviewer who asked you to gather up some pixie dust, I might get offended and consider it a breach of confidence.
Speaking generally, the reports are obviously a vital part of our scientific process, so it is strange that they so rarely see the light of day. Are there arguments within e.g. the open science movement to make all reviews publicly available?
A comment which is warranted, will be made however unjustified the authors may think the comment to be. For example, if you missed your commas or used wrong statistics, that will be pointed out or can be used as a cause for rejection both in closed reviews or open ones. You will get rejected because you don't come across as a professional person.
Open reviews protect authors from unwarranted comments such as gathering up pixie dust or doing 100 replicates of the same experiment.
At the same time, open reviews improve the article when peers from diverse fields provide their inputs, this leads to your growth as a scientist.
Also, I see a possibility for open review to hurt the peer-review process. When highly renowned authors submit articles for reviews containing very high stakes results, it makes a case for reviews to be closed. In an open review when the author has no peer of equal standing, reviewers may show restraint towards the author to protect their own interests, as the author may be a person who is reviewing their grant applications.
This may be an unpopular view, but academia just like life is an unfair place.